Re: devanagari question

[Jungshik Shin <jshin@xxxxxxxxxxx>]

On Sat, 3 Jan 2004 Andries.Brouwer@xxxxxx wrote:

>     From: Jungshik Shin <jshin@xxxxxxxxxxx>
>
> and all worked as it should. Excellent.
> (That is, the --enable-ctl --enable-xft works fine.

 Great.

> Have not tried the --enable-ctl version.)

  There's yet another method that just became the best of three although
it's just a _priliminary_ patch. I revised my patch for mozilla bug 215219
(http://bugzilla.mozilla.org/show_bug.cgi?id=215219).  With the patch,
Mozilla (--enable-xft, --enable-default-toolkit=gtk2, 'enable-ctl' is
NOT necessary [1]) takes advantage of Pango's complex script rendering
ability (http://www.pango.org) so that Mozilla's rendering capability
is only limited by Pango (in a sense). Devanagari, Malaylalam, Tamil,
Thai, Bengali, and more are supported with _standard_ opentype fonts
(as opposed to custom-encoded fonts). However, I have  more work to do
to support the cursor movement/selection at the  grapheme level. [2].
There are also issues with the code organization (some tasks done by
my patch had better be done in a different part of Mozilla). Besides,
I've been experiencing a mysterious crash at one particular URL (that
doesn't have anything but US-ASCII) with this patch.  Anyway, with my
latest patch to the bug, Mozilla-Xft (gtk2 + xft) is about on par with
Mozilla-Win on Windows __2k/XP__ in terms of complex script handling.

 With the patch,  you should list opentype Devanagari fonts such as
Raghindi at the beginning instead of SunIndic fonts.

  BTW,  IndicLinux seems to have distributed my old patch  for that bug
that has a trouble rendering 'justified' text. My latest patch solved
the problem with 'justified' text. (text-align: justify, letter-spacing:
..., word-spacing: .....)


> [By the way, --enable-ctl is documented as
> "Enable Thai Complex Script support". That should perhaps
>  be changed, as no Thai need be involved. I see
>
> so maybe Thai should be Thai/Devanagari.]

  Yes, currently, intl/ctl is for Thai and Devanagari.

> [The next step is vocalized Hebrew. Is the theory that it should
>  not work at all? What I get is a rendering with vowels in their
>  own space instead of on top / below the corresponding consonants.]

  You're right. That is, Hebrew vowels (vowel marks) are not rendered
as they should be.  I don't know whether Pango supports them. If you
have an opentype font supporting vocalized Hebrew, you may try 'gedit'
(one of many Gnome application that uses Pango) to see how it works.
You may contact Owen Taylor (otalyor redhat comm) or file a bug
at http://bugzilla.gnome.org (it vocalized Hebrew is not yet supported.)
See also bugs returned by the following query:

http://bugzilla.mozilla.org/buglist.cgi?product=Browser&component=Layout%3A+BiDi+Hebrew+%26+Arabic&bug_status=NEW&bug_status=ASSIGNED&bug_status=REOPENED

  At the moment, my patch mentioned above doesn't invoke Pango APIs for
Hebrew and Arabic (because they're internally handled by Mozilla. However,
Mozilla's internal BIDI handling routine doesn't handle vocalized Hebrew
as far as I know (I can be wrong). Simon Montagu < smontagu at smontagu org >
is the resident guru on Hebrew/Arabic issues in Mozilla)

  Jungshik

[1] With 'ctl + gtk2', I couldn't compile mozilla. This is a known bug
and a patch has been sitting somewhere.

[2] For the cursor movement and text selection, you may want
to apply the latest patch in bug 188288
( http://bugzilla.mozilla.org/show_bug.cgi?id=188288 ). Note that
this _only_ works for '--enable-ctl + --enable-xft' (ctl + gtk1 + xft)
or '--enable-ctl' (ctl + gtk1 + X11core) and does NOT work
with '--enable-xft --enable-default-toolkit=gtk2' with my latest
patch for bug 215219.


--
Linux-UTF8:   i18n of Linux on all levels
Archive:      http://mail.nl.linux.org/linux-utf8/